{"id":1869,"date":"2026-03-30T21:19:12","date_gmt":"2026-03-30T13:19:12","guid":{"rendered":"https:\/\/eve2333.top\/?p=1869"},"modified":"2026-03-30T21:19:12","modified_gmt":"2026-03-30T13:19:12","slug":"%e6%89%8b%e6%90%93rag%e7%b3%bb%e7%bb%9f-ragent-ai%ef%bc%883%ef%bc%89","status":"publish","type":"post","link":"https:\/\/eve2333.top\/?p=1869","title":{"rendered":"\u624b\u6413RAG\u7cfb\u7edf \u2013 Ragent AI\uff083\uff09"},"content":{"rendered":"\n<h1 class=\"wp-block-heading\">\u5143\u6570\u636e\u7684\u4f5c\u7528\u4e0e\u7ba1\u7406<\/h1>\n\n\n\n<p>\u5143\u6570\u636e\u662f\u5728\u5206\u5757\u4e4b\u540e\u3001\u5411\u91cf\u5316\u4e4b\u524d\u52a0\u5165\u7684\u3002\u5206\u5757\u5b8c\u6210\u540e\uff0c\u4f60\u5f97\u5230\u7684\u662f\u4e00\u4e2a\u4e2a\u7eaf\u6587\u672c\u5757\uff0c\u8fd9\u65f6\u5019\u7ed9\u6bcf\u4e2a\u5757\u6253\u4e0a\u6807\u7b7e\uff0c\u8bb0\u5f55\u5b83\u7684\u6765\u6e90\u3001\u6743\u9650\u3001\u4f4d\u7f6e\u7b49\u4fe1\u606f\u3002<\/p>\n\n\n\n<p>\u8fd9\u4e9b\u5143\u6570\u636e\u4f1a\u548c\u6587\u672c\u5185\u5bb9\u4e00\u8d77\uff0c\u88ab\u9001\u5230\u5411\u91cf\u6570\u636e\u5e93\u91cc\u5b58\u50a8\u3002\u540e\u7eed\u68c0\u7d22\u7684\u65f6\u5019\uff0c\u4e0d\u4ec5\u53ef\u4ee5\u6839\u636e\u6587\u672c\u76f8\u4f3c\u5ea6\u627e\u5230\u76f8\u5173\u7684\u5757\uff0c\u8fd8\u53ef\u4ee5\u6839\u636e\u5143\u6570\u636e\u505a\u8fc7\u6ee4\u3001\u6392\u5e8f\u3001\u5f15\u7528\u751f\u6210\u7b49\u64cd\u4f5c\u3002<\/p>\n\n\n\n<p>\u5728\u6ca1\u6709\u5143\u6570\u636e\u4e4b\u524d\uff0c\u4e00\u4e2a chunk \u5c31\u662f\u4e00\u6bb5\u6587\u672c\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>\"\u81ea\u7b7e\u6536\u4e4b\u65e5\u8d77 7 \u5929\u5185\uff0c\u5546\u54c1\u672a\u7ecf\u4f7f\u7528\u4e14\u4e0d\u5f71\u54cd\u4e8c\u6b21\u9500\u552e\u7684\uff0c\u6d88\u8d39\u8005\u53ef\u7533\u8bf7\u4e03\u5929\u65e0\u7406\u7531\u9000\u8d27\u3002\"<\/code><\/pre>\n\n\n\n<p>\u52a0\u4e0a\u5143\u6570\u636e\u4e4b\u540e\uff0c\u5b83\u53d8\u6210\u4e86\u8fd9\u6837\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>{\"content\":\"\u81ea\u7b7e\u6536\u4e4b\u65e5\u8d77 7 \u5929\u5185\uff0c\u5546\u54c1\u672a\u7ecf\u4f7f\u7528\u4e14\u4e0d\u5f71\u54cd\u4e8c\u6b21\u9500\u552e\u7684\uff0c\u6d88\u8d39\u8005\u53ef\u7533\u8bf7\u4e03\u5929\u65e0\u7406\u7531\u9000\u8d27\u3002\",\"metadata\":{\"doc_id\":\"doc_20240315_001\",\"source_url\":\"https:\/\/docs.company.com\/policy\/return.pdf\",\"file_name\":\"\u9000\u8d27\u653f\u7b56.pdf\",\"title\":\"\u4e00\u3001\u9000\u8d27\u653f\u7b56\",\"page_number\":3,\"created_at\":\"2024-03-15T10:30:00Z\",\"updated_at\":\"2024-03-15T10:30:00Z\",\"department\":\"customer_service\",\"access_roles\":&#91;\"employee\",\"customer_service\",\"manager\"],\"start_offset\":0,\"end_offset\":58,\"chunk_index\":0}}<\/code><\/pre>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"1-doc-id-source-url-file-name\">1. \u6587\u6863\u6807\u8bc6\u7c7b\uff1adoc_id\u3001source_url\u3001file_name<\/h3>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"1-1\">1.1 \u4e3a\u4ec0\u4e48\u9700\u8981\u6587\u6863\u6807\u8bc6<\/h4>\n\n\n\n<p>\u6700\u57fa\u7840\u7684\u9700\u6c42\uff1a\u77e5\u9053\u8fd9\u4e2a chunk \u6765\u81ea\u54ea\u4efd\u6587\u6863\u3002<\/p>\n\n\n\n<p>\u7528\u6237\u95ee\u201c\u62a5\u9500\u6d41\u7a0b\u662f\u4ec0\u4e48\u201d\uff0c\u7cfb\u7edf\u56de\u7b54\u540e\uff0c\u7528\u6237\u53ef\u80fd\u4f1a\u8ffd\u95ee\u201c\u80fd\u628a\u5b8c\u6574\u7684\u62a5\u9500\u6587\u6863\u53d1\u7ed9\u6211\u5417\u201d\u3002\u8fd9\u65f6\u5019\u5982\u679c chunk \u91cc\u8bb0\u5f55\u4e86 <code>source_url<\/code>\uff0c\u7cfb\u7edf\u5c31\u80fd\u76f4\u63a5\u8fd4\u56de\u6587\u6863\u94fe\u63a5\u3002<\/p>\n\n\n\n<p>\u53e6\u4e00\u4e2a\u573a\u666f\u662f\u53bb\u91cd\u548c\u66f4\u65b0\u3002\u5047\u8bbe\u4f60\u91cd\u65b0\u4e0a\u4f20\u4e86\u4e00\u4efd\u66f4\u65b0\u540e\u7684\u300a\u62a5\u9500\u6d41\u7a0b v2.0\u300b\uff0c\u7cfb\u7edf\u9700\u8981\u628a\u65e7\u7248\u672c\u7684\u6240\u6709 chunk \u5220\u6389\u3002\u5982\u679c\u6bcf\u4e2a chunk \u90fd\u8bb0\u5f55\u4e86 <code>doc_id<\/code>\uff0c\u5c31\u53ef\u4ee5\u6279\u91cf\u5220\u9664 <code>doc_id = \"\u62a5\u9500\u6d41\u7a0b v1.0\"<\/code> \u7684\u6240\u6709\u5757\u3002<\/p>\n\n\n\n<p>\u4e09\u4e2a\u5b57\u6bb5\u7684\u5206\u5de5\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><code>doc_id<\/code>\uff1a\u6587\u6863\u7684\u552f\u4e00\u6807\u8bc6\u7b26\uff0c\u901a\u5e38\u662f\u7cfb\u7edf\u751f\u6210\u7684 ID\uff0c\u7528\u4e8e\u5185\u90e8\u7ba1\u7406<\/li>\n\n\n\n<li><code>source_url<\/code>\uff1a\u6587\u6863\u7684\u8bbf\u95ee\u5730\u5740\uff0c\u53ef\u4ee5\u662f HTTP \u94fe\u63a5\u3001\u6587\u4ef6\u8def\u5f84\u3001\u6216\u8005\u5185\u90e8\u7cfb\u7edf\u7684 URL<\/li>\n\n\n\n<li><code>file_name<\/code>\uff1a\u6587\u6863\u7684\u539f\u59cb\u6587\u4ef6\u540d\uff0c\u65b9\u4fbf\u4eba\u7c7b\u9605\u8bfb<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"2-h1-h2-h3\">2. \u7ed3\u6784\u4fe1\u606f\u7c7b\uff1a\u6807\u9898\u5c42\u7ea7(H1\/H2\/H3)\u3001\u7ae0\u8282\u7f16\u53f7\u3001\u9875\u7801<\/h3>\n\n\n\n<p>\u5f88\u591a\u6587\u6863\u662f\u6709\u5c42\u7ea7\u7ed3\u6784\u7684\uff1a\u4e00\u7ea7\u6807\u9898\u3001\u4e8c\u7ea7\u6807\u9898\u3001\u4e09\u7ea7\u6807\u9898\u2026\u2026\u6bcf\u4e2a chunk \u5f80\u5f80\u5c5e\u4e8e\u67d0\u4e2a\u7ae0\u8282\u3002<\/p>\n\n\n\n<p>\u8bb0\u5f55\u7ed3\u6784\u4fe1\u606f\u6709\u4e24\u4e2a\u597d\u5904\uff1a<\/p>\n\n\n\n<p>\u4e00\u662f\u751f\u6210\u5f15\u7528\u65f6\u66f4\u7cbe\u786e\u3002\u6bd4\u5982\u7528\u6237\u95ee\u201c\u8bd5\u7528\u671f\u591a\u4e45\u201d\uff0c\u7cfb\u7edf\u56de\u7b54\u201c3 \u4e2a\u6708\u201d\uff0c\u5982\u679c chunk \u91cc\u8bb0\u5f55\u4e86 <code>title = \"\u7b2c\u4e8c\u7ae0 \u5458\u5de5\u5165\u804c &gt; 2.1 \u8bd5\u7528\u671f\u89c4\u5b9a\"<\/code>\uff0c\u5c31\u53ef\u4ee5\u8fd4\u56de\u201c\u4f9d\u636e\uff1a\u300a\u5458\u5de5\u624b\u518c\u300b\u7b2c\u4e8c\u7ae0 2.1 \u8282\u201d\u3002\u8fd9\u6bd4\u53ea\u8bf4\u201c\u6765\u6e90\uff1a\u5458\u5de5\u624b\u518c\u201d\u8981\u6e05\u6670\u5f97\u591a\u3002<\/p>\n\n\n\n<p>\u4e8c\u662f\u68c0\u7d22\u65f6\u53ef\u4ee5\u5229\u7528\u7ed3\u6784\u4fe1\u606f\u505a\u6392\u5e8f\u3002\u6bd4\u5982\u7528\u6237\u95ee\u7684\u662f\u4e00\u4e2a\u6982\u8ff0\u6027\u95ee\u9898\uff0c\u90a3\u4e00\u7ea7\u6807\u9898\u4e0b\u7684 chunk \u53ef\u80fd\u6bd4\u4e09\u7ea7\u6807\u9898\u4e0b\u7684\u66f4\u76f8\u5173\uff1b\u5982\u679c\u95ee\u7684\u662f\u7ec6\u8282\u95ee\u9898\uff0c\u5c31\u53cd\u8fc7\u6765\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"2-2\">2.2 \u5982\u4f55\u4ece\u539f\u59cb\u6587\u6863\u4e2d\u63d0\u53d6\u6807\u9898\u5c42\u7ea7<\/h4>\n\n\n\n<p>\u8fd9\u53d6\u51b3\u4e8e\u4f60\u7684\u6587\u6863\u683c\u5f0f\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Markdown \u6587\u6863<\/strong>\uff1a\u6807\u9898\u5c42\u7ea7\u662f\u73b0\u6210\u7684\uff08<code>#<\/code> \u662f H1\uff0c<code>##<\/code> \u662f H2\uff09<\/li>\n\n\n\n<li><strong>HTML \u6587\u6863<\/strong>\uff1a\u53ef\u4ee5\u89e3\u6790 <code>&lt;h1&gt;<\/code> <code>&lt;h2&gt;<\/code> <code>&lt;h3&gt;<\/code> \u6807\u7b7e<\/li>\n\n\n\n<li><strong>Word \u6587\u6863<\/strong>\uff1a\u53ef\u4ee5\u901a\u8fc7 Apache POI \u8bfb\u53d6\u6837\u5f0f\u4fe1\u606f\uff0c\u5224\u65ad\u54ea\u4e9b\u6bb5\u843d\u662f\u6807\u9898<\/li>\n\n\n\n<li><strong>PDF \u6587\u6863<\/strong>\uff1a\u6bd4\u8f83\u9ebb\u70e6\uff0c\u9700\u8981\u6839\u636e\u5b57\u4f53\u5927\u5c0f\u3001\u52a0\u7c97\u7b49\u7279\u5f81\u63a8\u65ad\u6807\u9898\uff0c\u6216\u8005\u7528\u4e13\u95e8\u7684 PDF \u89e3\u6790\u5e93<\/li>\n<\/ul>\n\n\n\n<p>\u4e00\u4e2a\u7b80\u5316\u7684\u601d\u8def\uff1a\u5728\u5206\u5757\u7684\u65f6\u5019\uff0c\u8bb0\u5f55\u5f53\u524d\u5757\u5c5e\u4e8e\u54ea\u4e2a\u6700\u8fd1\u7684\u6807\u9898\u3002\u6bd4\u5982\u4f60\u6309\u6bb5\u843d\u5207\u5757\uff0c\u5207\u5230\u7b2c 5 \u6bb5\u65f6\uff0c\u5f80\u524d\u627e\u6700\u8fd1\u7684\u4e00\u4e2a\u6807\u9898\u662f\u201c2.1 \u8bd5\u7528\u671f\u89c4\u5b9a\u201d\uff0c\u5c31\u628a\u8fd9\u4e2a\u6807\u9898\u8bb0\u5f55\u5230 chunk \u7684\u5143\u6570\u636e\u91cc\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"3\">3. \u65f6\u95f4\u7248\u672c\u7c7b\uff1a\u521b\u5efa\u65f6\u95f4\u3001\u66f4\u65b0\u65f6\u95f4\u3001\u751f\u6548\u65f6\u95f4\u3001\u5931\u6548\u65f6\u95f4<\/h3>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"3-1\">3.1 \u4e3a\u4ec0\u4e48\u9700\u8981\u65f6\u95f4\u7248\u672c\u4fe1\u606f<\/h4>\n\n\n\n<p>\u4f01\u4e1a\u7684\u77e5\u8bc6\u662f\u4f1a\u53d8\u5316\u7684\u3002\u4eca\u5e74\u7684\u62a5\u9500\u653f\u7b56\u548c\u53bb\u5e74\u4e0d\u4e00\u6837\uff0c\u4e0a\u4e2a\u6708\u7684\u4ea7\u54c1\u624b\u518c\u548c\u8fd9\u4e2a\u6708\u4e0d\u4e00\u6837\u3002\u5982\u679c\u4e0d\u8bb0\u5f55\u65f6\u95f4\u4fe1\u606f\uff0c\u7cfb\u7edf\u53ef\u80fd\u4f1a\u628a\u8fc7\u65f6\u7684\u5185\u5bb9\u8fd4\u56de\u7ed9\u7528\u6237\u3002<\/p>\n\n\n\n<p>\u56db\u4e2a\u65f6\u95f4\u5b57\u6bb5\u7684\u542b\u4e49\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><code>created_at<\/code>\uff1a\u8fd9\u4e2a chunk \u662f\u4ec0\u4e48\u65f6\u5019\u521b\u5efa\u7684\uff08\u901a\u5e38\u662f\u6587\u6863\u4e0a\u4f20\u5230\u7cfb\u7edf\u7684\u65f6\u95f4\uff09<\/li>\n\n\n\n<li><code>updated_at<\/code>\uff1a\u8fd9\u4e2a chunk \u6700\u540e\u4e00\u6b21\u66f4\u65b0\u662f\u4ec0\u4e48\u65f6\u5019<\/li>\n\n\n\n<li><code>effective_date<\/code>\uff1a\u8fd9\u6761\u89c4\u5219\u4ece\u4ec0\u4e48\u65f6\u5019\u5f00\u59cb\u751f\u6548\uff08\u6bd4\u5982\u201c\u65b0\u7248\u62a5\u9500\u653f\u7b56\u81ea 2024 \u5e74 4 \u6708 1 \u65e5\u8d77\u6267\u884c\u201d\uff09<\/li>\n\n\n\n<li><code>expiration_date<\/code>\uff1a\u8fd9\u6761\u89c4\u5219\u4ec0\u4e48\u65f6\u5019\u5931\u6548\uff08\u6bd4\u5982\u201c\u672c\u4f18\u60e0\u6d3b\u52a8\u622a\u6b62\u5230 2024 \u5e74 12 \u6708 31 \u65e5\u201d\uff09<\/li>\n<\/ul>\n\n\n\n<p>\u524d\u4e24\u4e2a\uff08<code>created_at<\/code>\u3001<code>updated_at<\/code>\uff09\u662f\u7cfb\u7edf\u81ea\u52a8\u8bb0\u5f55\u7684\uff0c\u540e\u4e24\u4e2a\uff08<code>effective_date<\/code>\u3001<code>expiration_date<\/code>\uff09\u901a\u5e38\u9700\u8981\u4ece\u6587\u6863\u5185\u5bb9\u4e2d\u63d0\u53d6\uff0c\u6216\u8005\u5728\u4e0a\u4f20\u6587\u6863\u65f6\u624b\u52a8\u6807\u6ce8\u3002<\/p>\n\n\n\n<p>\u4e00\u4e2a\u5178\u578b\u7684\u5e94\u7528\u573a\u666f\uff1a\u7528\u6237\u95ee\u201c\u73b0\u5728\u7684\u62a5\u9500\u989d\u5ea6\u662f\u591a\u5c11\u201d\uff0c\u7cfb\u7edf\u68c0\u7d22\u65f6\u53ef\u4ee5\u8fc7\u6ee4\u6389 <code>expiration_date<\/code> \u65e9\u4e8e\u5f53\u524d\u65e5\u671f\u7684 chunk\uff0c\u53ea\u8fd4\u56de\u5f53\u524d\u6709\u6548\u7684\u89c4\u5219\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"4-acl\">4. \u6743\u9650\u63a7\u5236\u7c7b\uff1a\u90e8\u95e8\u6807\u7b7e\u3001\u89d2\u8272\u6807\u7b7e\u3001ACL\uff08\u8bbf\u95ee\u63a7\u5236\u5217\u8868\uff09<\/h3>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"4-1-acl\">4.1 \u4ec0\u4e48\u662f ACL\uff0c\u4e3a\u4ec0\u4e48\u9700\u8981\u5b83<\/h4>\n\n\n\n<p>ACL \u662f Access Control List\uff08\u8bbf\u95ee\u63a7\u5236\u5217\u8868\uff09\u7684\u7f29\u5199\u3002\u7b80\u5355\u8bf4\uff0c\u5c31\u662f\u4e00\u4e2a\u5217\u8868\uff0c\u8bb0\u5f55\u4e86\u201c\u8c01\u80fd\u8bbf\u95ee\u8fd9\u4e2a\u8d44\u6e90\u201d\u3002<\/p>\n\n\n\n<p>\u5728 RAG \u7cfb\u7edf\u91cc\uff0c\u6bcf\u4e2a chunk \u53ef\u4ee5\u5e26\u4e00\u4e2a ACL\uff0c\u8bb0\u5f55\u54ea\u4e9b\u89d2\u8272\u3001\u54ea\u4e9b\u90e8\u95e8\u3001\u54ea\u4e9b\u7528\u6237\u80fd\u770b\u5230\u8fd9\u4e2a\u5757\u3002\u68c0\u7d22\u7684\u65f6\u5019\uff0c\u7cfb\u7edf\u4f1a\u5148\u68c0\u67e5\u5f53\u524d\u7528\u6237\u7684\u8eab\u4efd\uff0c\u7136\u540e\u53ea\u8fd4\u56de\u4ed6\u6709\u6743\u9650\u770b\u7684 chunk\u3002<\/p>\n\n\n\n<p>\u6743\u9650\u63a7\u5236\u6709\u51e0\u79cd\u5e38\u89c1\u7684\u7c92\u5ea6\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u57fa\u4e8e\u89d2\u8272<\/strong>\uff08Role-Based\uff09\uff1a\u6bd4\u5982 <code>access_roles: [\"manager\", \"hr\"]<\/code>\uff0c\u53ea\u6709\u7ecf\u7406\u548c HR \u80fd\u770b<\/li>\n\n\n\n<li><strong>\u57fa\u4e8e\u90e8\u95e8<\/strong>\uff08Department-Based\uff09\uff1a\u6bd4\u5982 <code>access_departments: [\"finance\", \"legal\"]<\/code>\uff0c\u53ea\u6709\u8d22\u52a1\u90e8\u548c\u6cd5\u52a1\u90e8\u80fd\u770b<\/li>\n\n\n\n<li><strong>\u57fa\u4e8e\u7528\u6237<\/strong>\uff08User-Based\uff09\uff1a\u6bd4\u5982 <code>access_users: [\"user_001\", \"user_002\"]<\/code>\uff0c\u53ea\u6709\u7279\u5b9a\u7528\u6237\u80fd\u770b<\/li>\n\n\n\n<li><strong>\u57fa\u4e8e\u654f\u611f\u7ea7\u522b<\/strong>\uff08Sensitivity-Based\uff09\uff1a\u6bd4\u5982 <code>sensitivity_level: \"public\"\/\"internal\"\/\"confidential\"\/\"secret\"<\/code>\uff0c\u4e0d\u540c\u7ea7\u522b\u7684\u7528\u6237\u80fd\u770b\u4e0d\u540c\u654f\u611f\u5ea6\u7684\u5185\u5bb9<\/li>\n<\/ul>\n\n\n\n<p>\u5b9e\u9645\u9879\u76ee\u4e2d\uff0c\u901a\u5e38\u4f1a\u7ec4\u5408\u4f7f\u7528\u3002\u6bd4\u5982\u4e00\u4e2a chunk \u6807\u8bb0\u4e3a <code>sensitivity_level: \"confidential\"<\/code>\uff0c\u540c\u65f6\u6307\u5b9a <code>access_departments: [\"finance\"]<\/code>\uff0c\u610f\u601d\u662f\u201c\u8fd9\u662f\u673a\u5bc6\u4fe1\u606f\uff0c\u53ea\u6709\u8d22\u52a1\u90e8\u80fd\u770b\u201d\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"4-3\">\u68c0\u7d22\u65f6\u5982\u4f55\u6839\u636e\u6743\u9650\u8fc7\u6ee4<\/h4>\n\n\n\n<p>\u6743\u9650\u8fc7\u6ee4\u901a\u5e38\u53d1\u751f\u5728\u68c0\u7d22\u9636\u6bb5\u3002\u5927\u591a\u6570\u5411\u91cf\u6570\u636e\u5e93\uff08\u5982 Milvus\u3001Qdrant\u3001Weaviate\uff09\u90fd\u652f\u6301\u5143\u6570\u636e\u8fc7\u6ee4\uff0c\u4f60\u53ef\u4ee5\u5728\u68c0\u7d22\u65f6\u52a0\u4e0a\u8fc7\u6ee4\u6761\u4ef6\u3002<\/p>\n\n\n\n<p>\u4f2a\u4ee3\u7801\u793a\u4f8b\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>\/\/ \u5047\u8bbe\u5f53\u524d\u7528\u6237\u662f\u8d22\u52a1\u90e8\u7684\u666e\u901a\u5458\u5de5String userRole =\"employee\";String userDepartment =\"finance\";\n\u200b\n\/\/ \u6784\u9020\u8fc7\u6ee4\u6761\u4ef6\uff1a\u53ea\u68c0\u7d22\u7528\u6237\u6709\u6743\u9650\u770b\u7684 chunkMap&lt;String, Object&gt; filter =newHashMap&lt;&gt;();\nfilter.put(\"access_roles\", userRole);\/\/ \u6216\u8005\nfilter.put(\"access_departments\", userDepartment);\n\u200b\n\/\/ \u6267\u884c\u68c0\u7d22\uff08\u5177\u4f53 API \u53d6\u51b3\u4e8e\u4f60\u7528\u7684\u5411\u91cf\u6570\u636e\u5e93\uff09List&lt;Document&gt; results = vectorStore.search(\n &nbsp; &nbsp;query,\n &nbsp; &nbsp;topK,\n &nbsp; &nbsp;filter &nbsp;\/\/ \u4f20\u5165\u8fc7\u6ee4\u6761\u4ef6);<\/code><\/pre>\n\n\n\n<p>\u4e0d\u540c\u5411\u91cf\u6570\u636e\u5e93\u7684\u8fc7\u6ee4\u8bed\u6cd5\u4e0d\u592a\u4e00\u6837\uff0c\u4f46\u6838\u5fc3\u601d\u8def\u662f\u4e00\u81f4\u7684\uff1a\u5728\u5411\u91cf\u76f8\u4f3c\u5ea6\u68c0\u7d22\u7684\u57fa\u7840\u4e0a\uff0c\u53e0\u52a0\u5143\u6570\u636e\u8fc7\u6ee4\u6761\u4ef6\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"5-start-offset-end-offset-chunk-index\">5. \u4f4d\u7f6e\u8ffd\u6eaf\u7c7b\uff1a\u539f\u6587\u4f4d\u7f6e\uff08start_offset\u3001end_offset\uff09\u3001chunk_index<\/h3>\n\n\n\n<p>\u5f53\u4f60\u53d1\u73b0\u67d0\u4e2a chunk \u7684\u5185\u5bb9\u6709\u95ee\u9898\uff0c\u9700\u8981\u56de\u5230\u539f\u6587\u53bb\u4fee\u6b63\u65f6\uff0c\u4f4d\u7f6e\u4fe1\u606f\u5c31\u6d3e\u4e0a\u7528\u573a\u4e86\u3002<\/p>\n\n\n\n<p><code>start_offset<\/code> \u548c <code>end_offset<\/code> \u8bb0\u5f55\u4e86\u8fd9\u4e2a chunk \u5728\u539f\u6587\u4e2d\u7684\u5b57\u7b26\u4f4d\u7f6e\u3002\u6bd4\u5982 <code>start_offset: 120, end_offset: 250<\/code>\uff0c\u610f\u601d\u662f\u8fd9\u4e2a chunk \u5bf9\u5e94\u539f\u6587\u7684\u7b2c 120 \u5230 250 \u4e2a\u5b57\u7b26\u3002<\/p>\n\n\n\n<p><code>chunk_index<\/code> \u662f\u8fd9\u4e2a chunk \u5728\u6240\u6709 chunk \u4e2d\u7684\u5e8f\u53f7\u3002\u6bd4\u5982\u4e00\u4efd\u6587\u6863\u88ab\u5207\u6210\u4e86 10 \u4e2a chunk\uff0c\u7b2c\u4e00\u4e2a chunk \u7684 <code>chunk_index<\/code> \u662f 0\uff0c\u7b2c\u4e8c\u4e2a\u662f 1\uff0c\u4ee5\u6b64\u7c7b\u63a8\u3002<\/p>\n\n\n\n<p>\u6709\u4e86\u8fd9\u4e9b\u4fe1\u606f\uff0c\u4f60\u53ef\u4ee5\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5feb\u901f\u5b9a\u4f4d\u5230\u539f\u6587\u7684\u5177\u4f53\u4f4d\u7f6e\uff0c\u65b9\u4fbf\u4eba\u5de5\u5ba1\u6838\u548c\u4fee\u6b63<\/li>\n\n\n\n<li>\u5728\u5c55\u793a\u7b54\u6848\u65f6\uff0c\u9ad8\u4eae\u663e\u793a\u539f\u6587\u4e2d\u7684\u76f8\u5173\u6bb5\u843d<\/li>\n\n\n\n<li>\u5206\u6790\u76f8\u90bb chunk \u7684\u5173\u7cfb\uff08\u6bd4\u5982\u68c0\u7d22\u5230\u4e86 chunk 5\uff0c\u53ef\u4ee5\u987a\u4fbf\u770b\u770b chunk 4 \u548c chunk 6 \u7684\u5185\u5bb9\uff09<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"6\">\u4e1a\u52a1\u81ea\u5b9a\u4e49\u7c7b\uff1a\u4ea7\u54c1\u7c7b\u578b\u3001\u4e1a\u52a1\u7ebf\u3001\u4f18\u5148\u7ea7\u7b49<\/h3>\n\n\n\n<p id=\"6-1\">\u4e1a\u52a1\u573a\u666f\u53ef\u80fd\u8fd8\u6709\u81ea\u5df1\u7279\u6b8a\u7684\u9700\u6c42\u3002<\/p>\n\n\n\n<p>\u6bd4\u5982\u7535\u5546\u573a\u666f\uff0c\u4f60\u53ef\u80fd\u9700\u8981\u7ed9 chunk \u6253\u4e0a\u5546\u54c1\u7c7b\u76ee\u6807\u7b7e\uff1a<\/p>\n\n\n\n<p>\u8fd9\u6837\u7528\u6237\u95ee\u201c\u751f\u9c9c\u80fd\u9000\u8d27\u5417\u201d\uff0c\u7cfb\u7edf\u53ef\u4ee5\u4f18\u5148\u68c0\u7d22 <code>product_category: \"fresh_food\"<\/code> \u7684 chunk\u3002<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u5143\u6570\u636e\u7684\u4e09\u5927\u6838\u5fc3\u5e94\u7528\u573a\u666f<\/h2>\n\n\n\n<p>\u524d\u9762\u8bb2\u4e86\u5143\u6570\u636e\u6709\u54ea\u4e9b\u5b57\u6bb5\uff0c\u73b0\u5728\u770b\u770b\u8fd9\u4e9b\u5b57\u6bb5\u5728\u5b9e\u9645\u573a\u666f\u4e2d\u600e\u4e48\u7528\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"1-ai\">1. \u56de\u7b54\u53ef\u5f15\u7528\uff1a\u8ba9 AI \u7684\u56de\u7b54\u6709\u636e\u53ef\u67e5<\/h3>\n\n\n\n<p>\u7528\u6237\u95ee\uff1a\u201c\u65b0\u5458\u5de5\u8bd5\u7528\u671f\u591a\u4e45\uff1f\u201d<\/p>\n\n\n\n<p>\u7cfb\u7edf\u56de\u7b54\uff1a\u201c\u65b0\u5458\u5de5\u8bd5\u7528\u671f\u4e3a 3 \u4e2a\u6708\u3002\u201d<\/p>\n\n\n\n<p>\u7528\u6237\u8ffd\u95ee\uff1a\u201c\u8fd9\u4e2a\u89c4\u5b9a\u5728\u54ea\u4efd\u6587\u6863\u91cc\uff1f\u201d<\/p>\n\n\n\n<p>\u5982\u679c chunk \u91cc\u8bb0\u5f55\u4e86\u6587\u6863\u6765\u6e90\u548c\u7ae0\u8282\u4fe1\u606f\uff0c\u7cfb\u7edf\u53ef\u4ee5\u8fd9\u6837\u56de\u7b54\uff1a<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>\u65b0\u5458\u5de5\u8bd5\u7528\u671f\u4e3a 3 \u4e2a\u6708\u3002<\/p>\n\n\n\n<p><strong>\u4f9d\u636e<\/strong>:\u300a\u5458\u5de5\u624b\u518c\u300b\u7b2c\u4e8c\u7ae0\u201c\u5458\u5de5\u5165\u804c\u201d &gt; 2.1 \u8282\u201c\u8bd5\u7528\u671f\u89c4\u5b9a\u201d\uff0c\u7b2c 5 \u9875<\/p>\n<\/blockquote>\n\n\n\n<p>\u8fd9\u5c31\u662f\u56de\u7b54\u53ef\u5f15\u7528\u3002\u7528\u6237\u4e0d\u4ec5\u5f97\u5230\u4e86\u7b54\u6848\uff0c\u8fd8\u77e5\u9053\u7b54\u6848\u7684\u51fa\u5904\uff0c\u53ef\u4ee5\u70b9\u51fb\u94fe\u63a5\u67e5\u770b\u5b8c\u6574\u7684\u539f\u6587\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"2\">2. \u6743\u9650\u8fc7\u6ee4\uff1a\u4e0d\u540c\u5458\u5de5\u770b\u4e0d\u540c\u77e5\u8bc6<\/h3>\n\n\n\n<p>\u516c\u53f8\u7684\u77e5\u8bc6\u5e93\u91cc\u6709\u5404\u79cd\u6587\u6863\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4ea7\u54c1\u90e8\u7684\u9700\u6c42\u6587\u6863\uff08\u53ea\u6709\u4ea7\u54c1\u90e8\u548c\u6280\u672f\u90e8\u80fd\u770b\uff09<\/li>\n\n\n\n<li>\u4eba\u4e8b\u90e8\u7684\u85aa\u916c\u653f\u7b56\uff08\u53ea\u6709\u4eba\u4e8b\u90e8\u548c\u7ba1\u7406\u5c42\u80fd\u770b\uff09<\/li>\n\n\n\n<li>\u8d22\u52a1\u90e8\u7684\u9884\u7b97\u6570\u636e\uff08\u53ea\u6709\u8d22\u52a1\u90e8\u548c\u9ad8\u7ba1\u80fd\u770b\uff09<\/li>\n\n\n\n<li>\u516c\u5171\u7684\u5458\u5de5\u624b\u518c\uff08\u6240\u6709\u4eba\u90fd\u80fd\u770b\uff09<\/li>\n<\/ul>\n\n\n\n<p>\u6280\u672f\u90e8\u7684\u5c0f\u674e\u95ee\uff1a\u201c\u516c\u53f8\u7684\u5e74\u7ec8\u5956\u600e\u4e48\u7b97\uff1f\u201d<\/p>\n\n\n\n<p>\u5982\u679c\u4e0d\u505a\u6743\u9650\u8fc7\u6ee4\uff0c\u7cfb\u7edf\u53ef\u80fd\u4f1a\u628a\u4eba\u4e8b\u90e8\u7684\u5185\u90e8\u6587\u6863\u8fd4\u56de\u7ed9\u4ed6\uff0c\u6cc4\u9732\u654f\u611f\u4fe1\u606f\u3002<\/p>\n\n\n\n<p>\u6b63\u786e\u7684\u505a\u6cd5\uff1a\u68c0\u7d22\u65f6\u6839\u636e\u5c0f\u674e\u7684\u8eab\u4efd\uff08\u6280\u672f\u90e8\u3001\u666e\u901a\u5458\u5de5\uff09\uff0c\u53ea\u8fd4\u56de\u4ed6\u6709\u6743\u9650\u770b\u7684 chunk\u3002\u5982\u679c\u6ca1\u6709\u5339\u914d\u7684\u516c\u5f00\u4fe1\u606f\uff0c\u5c31\u56de\u7b54\u201c\u62b1\u6b49\uff0c\u8fd9\u4e2a\u95ee\u9898\u6d89\u53ca\u5185\u90e8\u4fe1\u606f\uff0c\u8bf7\u54a8\u8be2\u4eba\u4e8b\u90e8\u201d\u3002<\/p>\n\n\n\n<p>\u68c0\u7d22\u524d\u6839\u636e\u7528\u6237\u89d2\u8272\u8fc7\u6ee4 chunk<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"3\">3.\u56de\u6eaf\u4e0e\u7ea0\u9519\uff1a\u53d1\u73b0\u9519\u7b54\u80fd\u5b9a\u4f4d\u5230\u6e90\u5934<\/h3>\n\n\n\n<p>\u7528\u6237\u53cd\u9988\uff1a\u201c\u7cfb\u7edf\u544a\u8bc9\u6211\u62a5\u9500\u9700\u8981\u8d34\u53d1\u7968\uff0c\u4f46\u5b9e\u9645\u4e0a\u65b0\u6d41\u7a0b\u5df2\u7ecf\u6539\u6210\u7ebf\u4e0a\u63d0\u4ea4\u4e86\uff0c\u4e0d\u7528\u8d34\u7eb8\u8d28\u53d1\u7968\u3002\u201d<\/p>\n\n\n\n<p>\u6280\u672f\u56e2\u961f\u9700\u8981\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>1.\u627e\u5230\u8fd4\u56de\u9519\u8bef\u4fe1\u606f\u7684\u90a3\u4e2a chunk<\/li>\n\n\n\n<li>2.\u5b9a\u4f4d\u5230\u539f\u59cb\u6587\u6863\u7684\u5177\u4f53\u4f4d\u7f6e<\/li>\n\n\n\n<li>3.\u4fee\u6b63\u6216\u5220\u9664\u8fd9\u4e2a chunk<\/li>\n\n\n\n<li>4.\u68c0\u67e5\u662f\u5426\u8fd8\u6709\u5176\u4ed6\u76f8\u5173\u7684\u8fc7\u65f6 chunk \u9700\u8981\u4e00\u8d77\u66f4\u65b0<\/li>\n<\/ol>\n\n\n\n<p>\u5982\u679c chunk \u91cc\u8bb0\u5f55\u4e86 <code>doc_id<\/code>\u3001<code>chunk_index<\/code>\u3001<code>start_offset<\/code>\u3001<code>created_at<\/code> \u7b49\u4fe1\u606f\uff0c\u8fd9\u4e2a\u8fc7\u7a0b\u5c31\u4f1a\u7b80\u5355\u5f88\u591a\u3002<\/p>\n\n\n\n<p id=\"3-2-chunk\">\u5b9e\u73b0\u601d\u8def\uff1a\u901a\u8fc7\u5143\u6570\u636e\u5b9a\u4f4d\u95ee\u9898 chunk\uff0c\u5feb\u901f\u5b9a\u4f4d\uff0c\u7cbe\u51c6\u4fee\u6b63\uff0c\u6279\u91cf\u66f4\u65b0\uff0c\u7248\u672c\u7ba1\u7406<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u5143\u6570\u636e\u8bbe\u8ba1\u7684\u6700\u4f73\u5b9e\u8df5<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"1\">1. \u5143\u6570\u636e\u5b57\u6bb5\u4e0d\u662f\u8d8a\u591a\u8d8a\u597d<\/h3>\n\n\n\n<p>\u95ee\u9898\u5728\u4e8e\uff1a<strong>\u5b58\u50a8\u6210\u672c\uff0c\u7ef4\u62a4\u6210\u672c\uff0c\u68c0\u7d22\u6027\u80fd<\/strong><\/p>\n\n\n\n<p>\u4e00\u4e2a\u5b9e\u7528\u7684\u539f\u5219\uff1a<strong>\u53ea\u52a0\u5bf9\u68c0\u7d22\u3001\u8fc7\u6ee4\u3001\u5c55\u793a\u6709\u5b9e\u9645\u5e2e\u52a9\u7684\u5b57\u6bb5<\/strong>\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"2\">2. \u5143\u6570\u636e\u7684\u7c92\u5ea6\u8981\u548c\u4e1a\u52a1\u573a\u666f\u5339\u914d<\/h3>\n\n\n\n<p>\u4e0d\u540c\u7684\u4e1a\u52a1\u573a\u666f\uff0c\u5bf9\u5143\u6570\u636e\u7684\u7c92\u5ea6\u8981\u6c42\u4e0d\u4e00\u6837\u3002<\/p>\n\n\n\n<p><strong>\u573a\u666f 1\uff1a\u9762\u5411\u516c\u4f17\u7684\u4ea7\u54c1\u5e2e\u52a9\u6587\u6863<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4e0d\u9700\u8981\u6743\u9650\u63a7\u5236\uff08\u6240\u6709\u4eba\u90fd\u80fd\u770b\uff09<\/li>\n\n\n\n<li>\u4e0d\u9700\u8981\u90e8\u95e8\u6807\u7b7e\uff08\u6ca1\u6709\u90e8\u95e8\u6982\u5ff5\uff09<\/li>\n\n\n\n<li>\u9700\u8981\u6587\u6863\u6807\u8bc6\u548c\u7ae0\u8282\u4fe1\u606f\uff08\u65b9\u4fbf\u5f15\u7528\uff09<\/li>\n<\/ul>\n\n\n\n<p>\u63a8\u8350\u7684\u5143\u6570\u636e\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>{\"doc_id\":\"...\",\"file_name\":\"...\",\"title\":\"...\",\"source_url\":\"...\"}<\/code><\/pre>\n\n\n\n<p><strong>\u573a\u666f 2\uff1a\u4f01\u4e1a\u5185\u90e8\u77e5\u8bc6\u5e93<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u9700\u8981\u6743\u9650\u63a7\u5236\uff08\u4e0d\u540c\u90e8\u95e8\u770b\u4e0d\u540c\u5185\u5bb9\uff09<\/li>\n\n\n\n<li>\u9700\u8981\u65f6\u95f4\u7248\u672c\uff08\u77e5\u8bc6\u4f1a\u66f4\u65b0\uff09<\/li>\n\n\n\n<li>\u9700\u8981\u4f4d\u7f6e\u8ffd\u6eaf\uff08\u65b9\u4fbf\u7ea0\u9519\uff09<\/li>\n<\/ul>\n\n\n\n<p>\u63a8\u8350\u7684\u5143\u6570\u636e\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>{\"doc_id\":\"...\",\"file_name\":\"...\",\"title\":\"...\",\"source_url\":\"...\",\"access_departments\":&#91;...],\"access_roles\":&#91;...],\"created_at\":\"...\",\"updated_at\":\"...\",\"start_offset\": ...,\"end_offset\": ...,\"chunk_index\": ...\n}<\/code><\/pre>\n\n\n\n<p><strong>\u573a\u666f 3\uff1a\u7535\u5546\u5ba2\u670d\u77e5\u8bc6\u5e93<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u9700\u8981\u5546\u54c1\u7c7b\u76ee\u6807\u7b7e\uff08\u4e0d\u540c\u7c7b\u76ee\u7684\u89c4\u5219\u4e0d\u540c\uff09<\/li>\n\n\n\n<li>\u9700\u8981\u653f\u7b56\u7c7b\u578b\u6807\u7b7e\uff08\u9000\u8d27\u3001\u6362\u8d27\u3001\u7269\u6d41\u7b49\uff09<\/li>\n\n\n\n<li>\u9700\u8981\u4f18\u5148\u7ea7\uff08\u67d0\u4e9b\u89c4\u5219\u4f18\u5148\u7ea7\u66f4\u9ad8\uff09<\/li>\n<\/ul>\n\n\n\n<p>\u63a8\u8350\u7684\u5143\u6570\u636e\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>{\"doc_id\":\"...\",\"file_name\":\"...\",\"product_category\":\"...\",\"policy_type\":\"...\",\"priority\": ...,\"effective_date\":\"...\",\"expiration_date\":\"...\"}<\/code><\/pre>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"3\">3. \u5143\u6570\u636e\u7684\u7ef4\u62a4\u6210\u672c\u8981\u8003\u8651\u8fdb\u53bb<\/h3>\n\n\n\n<p>\u6709\u4e9b\u5143\u6570\u636e\u662f\u7cfb\u7edf\u81ea\u52a8\u751f\u6210\u7684\uff08\u5982 <code>created_at<\/code>\u3001<code>chunk_index<\/code>\u3001<code>start_offset<\/code>\uff09\uff0c\u7ef4\u62a4\u6210\u672c\u4f4e\u3002<\/p>\n\n\n\n<p>\u6709\u4e9b\u5143\u6570\u636e\u9700\u8981\u4eba\u5de5\u6807\u6ce8\uff08\u5982 <code>access_roles<\/code>\u3001<code>product_category<\/code>\u3001<code>effective_date<\/code>\uff09\uff0c\u7ef4\u62a4\u6210\u672c\u9ad8\u3002<\/p>\n\n\n\n<p>\u5982\u679c\u4f60\u7684\u77e5\u8bc6\u5e93\u6709\u51e0\u5343\u4efd\u6587\u6863\uff0c\u6bcf\u4efd\u6587\u6863\u90fd\u8981\u4eba\u5de5\u6807\u6ce8\u5341\u51e0\u4e2a\u5b57\u6bb5\uff0c\u8fd9\u4e2a\u5de5\u4f5c\u91cf\u662f\u4e0d\u73b0\u5b9e\u7684\u3002<\/p>\n\n\n\n<p>\u4e00\u4e2a\u6298\u4e2d\u7684\u65b9\u6848\uff1a<strong>\u5206\u5c42\u6807\u6ce8<\/strong>\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u6587\u6863\u7ea7\u6807\u6ce8<\/strong>\uff1a\u5728\u4e0a\u4f20\u6587\u6863\u65f6\uff0c\u6807\u6ce8\u6587\u6863\u7ea7\u7684\u5143\u6570\u636e\uff08\u5982 <code>access_departments<\/code>\u3001<code>doc_type<\/code>\uff09\uff0c\u8fd9\u4e9b\u5143\u6570\u636e\u4f1a\u81ea\u52a8\u7ee7\u627f\u7ed9\u6587\u6863\u4e0b\u7684\u6240\u6709 chunk<\/li>\n\n\n\n<li><strong>Chunk \u7ea7\u6807\u6ce8<\/strong>\uff1a\u7cfb\u7edf\u81ea\u52a8\u751f\u6210 chunk \u7ea7\u7684\u5143\u6570\u636e\uff08\u5982 <code>chunk_index<\/code>\u3001<code>start_offset<\/code>\uff09<\/li>\n\n\n\n<li><strong>\u6309\u9700\u6807\u6ce8<\/strong>\uff1a\u53ea\u5bf9\u91cd\u8981\u7684\u3001\u9ad8\u9891\u8bbf\u95ee\u7684\u6587\u6863\u505a\u7cbe\u7ec6\u5316\u6807\u6ce8\uff08\u5982 <code>effective_date<\/code>\u3001<code>priority<\/code>\uff09<\/li>\n<\/ul>\n\n\n\n<p>\u8fd9\u6837\u53ef\u4ee5\u5728\u4fdd\u8bc1\u5143\u6570\u636e\u8d28\u91cf\u7684\u540c\u65f6\uff0c\u628a\u7ef4\u62a4\u6210\u672c\u63a7\u5236\u5728\u53ef\u63a5\u53d7\u7684\u8303\u56f4\u5185\u3002<\/p>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><tbody><tr><th class=\"has-text-align-left\" data-align=\"left\">\u5143\u6570\u636e\u5b57\u6bb5<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u7528\u9014<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u9002\u7528\u573a\u666f<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u7ef4\u62a4\u6210\u672c<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u4f18\u5148\u7ea7<\/th><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">doc_id<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u6587\u6863\u6807\u8bc6\uff0c\u7528\u4e8e\u6279\u91cf\u7ba1\u7406<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u51e0\u4e4e\u6240\u6709\u573a\u666f<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f4e\uff08\u7cfb\u7edf\u751f\u6210\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5fc5\u987b<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">file_name<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5c55\u793a\u7ed9\u7528\u6237\uff0c\u751f\u6210\u5f15\u7528<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u51e0\u4e4e\u6240\u6709\u573a\u666f<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f4e\uff08\u7cfb\u7edf\u751f\u6210\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5fc5\u987b<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">source_url<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u63d0\u4f9b\u539f\u6587\u94fe\u63a5<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u9700\u8981\u56de\u6eaf\u539f\u6587\u7684\u573a\u666f<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f4e\uff08\u7cfb\u7edf\u751f\u6210\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u63a8\u8350<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">title \/ h1_title \/ h2_title<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u751f\u6210\u5f15\u7528\uff0c\u5c55\u793a\u7ae0\u8282\u4fe1\u606f<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u6709\u7ed3\u6784\u7684\u6587\u6863<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4e2d\uff08\u9700\u8981\u89e3\u6790\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u63a8\u8350<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">page_number<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u751f\u6210\u5f15\u7528\uff0c\u5b9a\u4f4d\u539f\u6587<\/td><td class=\"has-text-align-left\" data-align=\"left\">PDF \u7b49\u6709\u9875\u7801\u7684\u6587\u6863<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4e2d\uff08\u9700\u8981\u89e3\u6790\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u63a8\u8350<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">created_at \/ updated_at<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u7248\u672c\u7ba1\u7406\uff0c\u8ffd\u8e2a\u53d8\u66f4<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u77e5\u8bc6\u4f1a\u66f4\u65b0\u7684\u573a\u666f<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f4e\uff08\u7cfb\u7edf\u751f\u6210\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u63a8\u8350<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">effective_date \/ expiration_date<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u8fc7\u6ee4\u8fc7\u65f6\u5185\u5bb9<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u6709\u65f6\u6548\u6027\u7684\u77e5\u8bc6\uff08\u653f\u7b56\u3001\u6d3b\u52a8\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u9ad8\uff08\u9700\u8981\u4eba\u5de5\u6807\u6ce8\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u53ef\u9009<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">access_roles \/ access_departments<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u6743\u9650\u8fc7\u6ee4<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f01\u4e1a\u5185\u90e8\u77e5\u8bc6\u5e93<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u9ad8\uff08\u9700\u8981\u4eba\u5de5\u6807\u6ce8\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5fc5\u987b\uff08\u5982\u679c\u6709\u6743\u9650\u9700\u6c42\uff09<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">sensitivity_level<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u6743\u9650\u8fc7\u6ee4<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f01\u4e1a\u5185\u90e8\u77e5\u8bc6\u5e93<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4e2d\uff08\u53ef\u4ee5\u6309\u6587\u6863\u6807\u6ce8\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u63a8\u8350\uff08\u5982\u679c\u6709\u6743\u9650\u9700\u6c42\uff09<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">start_offset \/ end_offset<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5b9a\u4f4d\u539f\u6587\uff0c\u7ea0\u9519<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u9700\u8981\u4eba\u5de5\u5ba1\u6838\u548c\u4fee\u6b63\u7684\u573a\u666f<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f4e\uff08\u7cfb\u7edf\u751f\u6210\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u63a8\u8350<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">chunk_index<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5b9a\u4f4d chunk\uff0c\u5206\u6790\u76f8\u90bb\u5757<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u9700\u8981\u7ba1\u7406 chunk \u7684\u573a\u666f<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f4e\uff08\u7cfb\u7edf\u751f\u6210\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u63a8\u8350<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">product_category \/ policy_type \u7b49<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4e1a\u52a1\u8fc7\u6ee4\u548c\u6392\u5e8f<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u6709\u660e\u786e\u4e1a\u52a1\u5206\u7c7b\u7684\u573a\u666f<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4e2d\u5230\u9ad8\uff08\u53d6\u51b3\u4e8e\u5206\u7c7b\u590d\u6742\u5ea6\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u53ef\u9009<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<p>\u4f7f\u7528\u5efa\u8bae\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u5fc5\u987b<\/strong>:\u8fd9\u4e9b\u5b57\u6bb5\u51e0\u4e4e\u6240\u6709\u573a\u666f\u90fd\u9700\u8981\uff0c\u4f18\u5148\u5b9e\u73b0<\/li>\n\n\n\n<li><strong>\u63a8\u8350<\/strong>:\u8fd9\u4e9b\u5b57\u6bb5\u80fd\u663e\u8457\u63d0\u5347\u7528\u6237\u4f53\u9a8c\u6216\u8fd0\u7ef4\u6548\u7387\uff0c\u5efa\u8bae\u5b9e\u73b0<\/li>\n\n\n\n<li><strong>\u53ef\u9009<\/strong>:\u8fd9\u4e9b\u5b57\u6bb5\u9488\u5bf9\u7279\u5b9a\u573a\u666f\uff0c\u6839\u636e\u5b9e\u9645\u9700\u6c42\u51b3\u5b9a\u662f\u5426\u5b9e\u73b0<\/li>\n<\/ul>\n\n\n\n<h1 class=\"wp-block-heading\">\u4ece\u6587\u672c\u5230\u5411\u91cf\u4e4b\u7406\u89e3Embedding<\/h1>\n\n\n\n<h4 class=\"wp-block-heading\">2.1 \u540c\u4e49\u8bcd\u95ee\u9898<\/h4>\n\n\n\n<p>\u201c\u4e00\u5468\u201d\u548c\u201c\u4e03\u5929\u201d\u662f\u540c\u4e00\u4e2a\u610f\u601d\uff0c\u201c\u9000\u201d\u548c\u201c\u9000\u8d27\u201d\u662f\u540c\u4e00\u4e2a\u52a8\u4f5c\uff0c\u4f46\u5173\u952e\u8bcd\u68c0\u7d22\u4e0d\u77e5\u9053\u3002\u5b83\u53ea\u505a\u5b57\u9762\u5339\u914d\uff0c\u4e0d\u7406\u89e3\u8bed\u4e49\u3002\u65e5\u5e38\u5bf9\u8bdd\u4e2d\u7684\u8868\u8fbe\u5173\u952e\u8bcd\u68c0\u7d22\u4e0d\u80fd\uff0c\u56e0\u4e3a\u4e66\u9762<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"2-2\">2.2 \u4e00\u8bcd\u591a\u4e49\u95ee\u9898<\/h4>\n\n\n\n<p>\u540c\u4e00\u4e2a\u8bcd\u5728\u4e0d\u540c\u8bed\u5883\u4e0b\u610f\u601d\u5b8c\u5168\u4e0d\u540c\u3002\u7528\u6237\u95ee\u201c\u82f9\u679c\u7684\u552e\u540e\u653f\u7b56\u662f\u4ec0\u4e48\u201d\u2014\u2014\u8fd9\u91cc\u7684\u201c\u82f9\u679c\u201d\u662f\u6307 Apple \u54c1\u724c\uff0c\u8fd8\u662f\u6307\u6c34\u679c\uff1f\u5173\u952e\u8bcd\u68c0\u7d22\u4e0d\u77e5\u9053\uff0c\u5b83\u4f1a\u628a\u6240\u6709\u5305\u542b\u201c\u82f9\u679c\u201d\u7684\u6587\u672c\u5757\u90fd\u8fd4\u56de\uff0c\u6c34\u679c\u7c7b\u76ee\u7684\u9000\u8d27\u653f\u7b56\u548c Apple \u4ea7\u54c1\u7684\u4fdd\u4fee\u653f\u7b56\u6df7\u5728\u4e00\u8d77\u3002<\/p>\n\n\n\n<p>\u518d\u6bd4\u5982\u201c\u5145\u503c\u201d\u8fd9\u4e2a\u8bcd\uff0c\u5728\u6e38\u620f\u573a\u666f\u662f\u5145\u6e38\u620f\u5e01\uff0c\u5728\u8bdd\u8d39\u573a\u666f\u662f\u5145\u8bdd\u8d39\uff0c\u5728\u4f1a\u5458\u573a\u666f\u662f\u5145\u4f1a\u5458\u4f59\u989d\u3002\u5173\u952e\u8bcd\u68c0\u7d22\u6ca1\u6cd5\u533a\u5206\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"2-3\">2.3 \u4e0a\u4e0b\u6587\u7406\u89e3\u95ee\u9898<\/h4>\n\n\n\n<p>\u6709\u4e9b\u95ee\u9898\u9700\u8981\u7406\u89e3\u6574\u53e5\u8bdd\u7684\u610f\u601d\uff0c\u800c\u4e0d\u662f\u62c6\u6210\u5355\u4e2a\u5173\u952e\u8bcd\u3002<\/p>\n\n\n\n<p>\u7528\u6237\u95ee\uff1a\u201c\u6211\u4e0d\u60f3\u8981\u4e86\uff0c\u4f46\u5df2\u7ecf\u62c6\u4e86\u5305\u88c5\u3002\u201d<\/p>\n\n\n\n<p>\u5173\u952e\u8bcd\u68c0\u7d22\u62c6\u51fa\u6765\u7684\u662f\u201c\u4e0d\u60f3\u8981\u201d\u201c\u62c6\u4e86\u201d\u201c\u5305\u88c5\u201d\uff0c\u53ef\u80fd\u5339\u914d\u5230\u201c\u5305\u88c5\u6750\u6599\u8bf4\u660e\u201d\u6216\u8005\u201c\u62c6\u7bb1\u6307\u5357\u201d\u8fd9\u79cd\u5b8c\u5168\u4e0d\u76f8\u5173\u7684\u5185\u5bb9\u3002\u4f46\u5b9e\u9645\u4e0a\u7528\u6237\u60f3\u95ee\u7684\u662f\u201c\u62c6\u5c01\u5546\u54c1\u80fd\u4e0d\u80fd\u9000\u8d27\u201d\u3002<\/p>\n\n\n\n<p>\u7528\u4e00\u53e5\u8bdd\u6982\u62ec Embedding \u7684\u6838\u5fc3\u601d\u60f3\uff1a\u628a\u6587\u672c\u6620\u5c04\u5230\u4e00\u4e2a\u9ad8\u7ef4\u7a7a\u95f4\u4e2d\uff0c\u8ba9\u8bed\u4e49\u76f8\u8fd1\u7684\u6587\u672c\u5728\u7a7a\u95f4\u4e2d\u8ddd\u79bb\u76f8\u8fd1\u3002<\/p>\n\n\n\n<p>\u771f\u5b9e\u7684\u8bed\u8a00\u542b\u4e49\u975e\u5e38\u4e30\u5bcc\uff0c\u4e24\u4e2a\u7ef4\u5ea6\u6839\u672c\u4e0d\u591f\u7528\u3002<\/p>\n\n\n\n<p>\u5b9e\u9645\u7684 Embedding \u6a21\u578b\u4f1a\u7528\u51e0\u767e\u751a\u81f3\u4e0a\u5343\u4e2a\u7ef4\u5ea6\u6765\u8868\u793a\u4e00\u6bb5\u6587\u672c\u3002\u6bcf\u4e2a\u7ef4\u5ea6\u6355\u6349\u6587\u672c\u542b\u4e49\u7684\u4e00\u4e2a\u65b9\u9762\u2014\u2014\u867d\u7136\u6211\u4eec\u6ca1\u6cd5\u76f4\u89c2\u5730\u8bf4\u51fa\u6bcf\u4e2a\u7ef4\u5ea6\u5177\u4f53\u4ee3\u8868\u4ec0\u4e48\uff0c\u4f46\u6a21\u578b\u901a\u8fc7\u5927\u91cf\u8bad\u7ec3\u6570\u636e\u5b66\u4f1a\u4e86\u600e\u4e48\u5206\u914d\u8fd9\u4e9b\u7ef4\u5ea6\u3002<\/p>\n\n\n\n<p><\/p>\n\n\n\n<p>\u56de\u5230\u5f00\u5934\u7684\u4f8b\u5b50\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u201c\u4e03\u5929\u65e0\u7406\u7531\u9000\u8d27\u201d\u7684\u5411\u91cf\u548c\u201c\u4e70\u4e86\u4e00\u5468\u7684\u4e1c\u897f\u8fd8\u80fd\u9000\u5417\u201d\u7684\u5411\u91cf\uff0c\u5728\u9ad8\u7ef4\u7a7a\u95f4\u4e2d\u8ddd\u79bb\u5f88\u8fd1<\/li>\n\n\n\n<li>\u201c\u4e03\u5929\u65e0\u7406\u7531\u9000\u8d27\u201d\u7684\u5411\u91cf\u548c\u201c\u7269\u6d41\u914d\u9001\u65f6\u6548\u8bf4\u660e\u201d\u7684\u5411\u91cf\uff0c\u8ddd\u79bb\u5f88\u8fdc<\/li>\n<\/ul>\n\n\n\n<p>\u6709\u4e86\u5411\u91cf\u8868\u793a\uff0c\u8ba1\u7b97\u673a\u5c31\u4e0d\u7528\u518d\u505a\u5b57\u9762\u5339\u914d\u4e86\u3002\u5b83\u53ea\u9700\u8981\u6bd4\u8f83\u4e24\u4e2a\u5411\u91cf\u4e4b\u95f4\u7684\u8ddd\u79bb\uff0c\u8ddd\u79bb\u8fd1\u5c31\u662f\u8bed\u4e49\u76f8\u5173\uff0c\u8ddd\u79bb\u8fdc\u5c31\u662f\u8bed\u4e49\u65e0\u5173\u3002<\/p>\n\n\n\n<p>\u8fd9\u5c31\u662f RAG \u7cfb\u7edf\u80fd\u505a\u8bed\u4e49\u68c0\u7d22\u7684\u6839\u57fa\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u51e0\u4e2a\u5173\u952e\u7279\u6027\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u8f93\u5165\u957f\u5ea6\u6709\u9650\u5236\uff1a\u6bcf\u4e2a\u6a21\u578b\u90fd\u6709\u6700\u5927\u8f93\u5165 token \u6570\uff08\u6bd4\u5982 512 \u6216 8192 \u7b49\u7b49\uff09\uff0c\u8d85\u8fc7\u4f1a\u88ab\u622a\u65ad\u3002\u8fd9\u4e5f\u662f\u4e3a\u4ec0\u4e48\u524d\u9762\u8981\u505a\u5206\u5757\u2014\u2014\u628a\u957f\u6587\u6863\u5207\u6210\u77ed\u6587\u672c\u5757\uff0c\u786e\u4fdd\u6bcf\u4e2a\u5757\u90fd\u5728\u6a21\u578b\u7684\u8f93\u5165\u9650\u5236\u5185<\/li>\n\n\n\n<li>\u8f93\u51fa\u7ef4\u5ea6\u56fa\u5b9a\uff1a\u540c\u4e00\u4e2a\u6a21\u578b\u8f93\u51fa\u7684\u5411\u91cf\u7ef4\u5ea6\u662f\u56fa\u5b9a\u7684\u3002\u6bd4\u5982\u67d0\u4e2a\u6a21\u578b\u8f93\u51fa 1024 \u7ef4\uff0c\u90a3\u4e0d\u7ba1\u4f60\u8f93\u5165\u4e00\u4e2a\u8bcd\u8fd8\u662f\u4e00\u6bb5\u8bdd\uff0c\u8f93\u51fa\u90fd\u662f 1024 \u4e2a\u6d6e\u70b9\u6570<\/li>\n\n\n\n<li>\u540c\u4e00\u6a21\u578b\u5185\u53ef\u6bd4\u8f83\uff1a\u53ea\u6709\u7528\u540c\u4e00\u4e2a\u6a21\u578b\u751f\u6210\u7684\u5411\u91cf\u624d\u80fd\u4e92\u76f8\u6bd4\u8f83\u3002\u7528\u6a21\u578b A \u751f\u6210\u7684\u5411\u91cf\u548c\u6a21\u578b B \u751f\u6210\u7684\u5411\u91cf\uff0c\u6ca1\u6cd5\u76f4\u63a5\u7b97\u76f8\u4f3c\u5ea6<\/li>\n<\/ul>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>\u7b2c\u4e09\u70b9\u975e\u5e38\u91cd\u8981\u3002\u8fd9\u610f\u5473\u7740\u4f60\u5728\u6570\u636e\u51c6\u5907\u9636\u6bb5\u7528\u4ec0\u4e48\u6a21\u578b\u628a chunk \u8f6c\u6210\u5411\u91cf\uff0c\u68c0\u7d22\u9636\u6bb5\u5c31\u5fc5\u987b\u7528\u540c\u4e00\u4e2a\u6a21\u578b\u628a\u7528\u6237\u7684 query \u8f6c\u6210\u5411\u91cf\u3002<strong>\u6362\u6a21\u578b = \u6240\u6709\u5411\u91cf\u8981\u91cd\u65b0\u751f\u6210<\/strong>\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>\u5728\u770b\u5177\u4f53\u6a21\u578b\u4e4b\u524d\uff0c\u5148\u641e\u6e05\u695a\u9009\u578b\u65f6\u8981\u770b\u54ea\u4e9b\u6307\u6807\uff1a<\/p>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><tbody><tr><th class=\"has-text-align-left\" data-align=\"left\">\u6307\u6807<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u542b\u4e49<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u4e3a\u4ec0\u4e48\u91cd\u8981<\/th><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">\u5411\u91cf\u7ef4\u5ea6<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u8f93\u51fa\u5411\u91cf\u7684\u6d6e\u70b9\u6570\u4e2a\u6570<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u7ef4\u5ea6\u8d8a\u9ad8\uff0c\u8868\u8fbe\u80fd\u529b\u8d8a\u5f3a\uff0c\u4f46\u5b58\u50a8\u548c\u8ba1\u7b97\u6210\u672c\u4e5f\u8d8a\u9ad8<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">\u6700\u5927\u8f93\u5165 token \u6570<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5355\u6b21\u80fd\u5904\u7406\u7684\u6700\u5927\u6587\u672c\u957f\u5ea6<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u51b3\u5b9a\u4e86\u4f60\u7684 chunk \u6700\u5927\u80fd\u6709\u591a\u957f<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">\u4e2d\u6587\u6548\u679c<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5bf9\u4e2d\u6587\u6587\u672c\u7684\u8bed\u4e49\u7406\u89e3\u80fd\u529b<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4e2d\u6587\u573a\u666f\u5fc5\u987b\u5173\u6ce8\uff0c\u6709\u4e9b\u6a21\u578b\u4e3b\u8981\u9488\u5bf9\u82f1\u6587\u8bad\u7ec3<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">API \u6210\u672c<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u6bcf\u6b21\u8c03\u7528\u7684\u8d39\u7528<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5927\u89c4\u6a21\u5411\u91cf\u5316\u65f6\uff0c\u6210\u672c\u5dee\u5f02\u4f1a\u5f88\u660e\u663e<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">\u662f\u5426\u652f\u6301\u672c\u5730\u90e8\u7f72<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u80fd\u4e0d\u80fd\u5728\u81ea\u5df1\u7684\u670d\u52a1\u5668\u4e0a\u8dd1<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u6d89\u53ca\u6570\u636e\u5b89\u5168\u548c\u9690\u79c1\u7684\u573a\u666f\uff0c\u53ef\u80fd\u4e0d\u5141\u8bb8\u6570\u636e\u51fa\u5916\u7f51<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><tbody><tr><th class=\"has-text-align-left\" data-align=\"left\">\u6a21\u578b<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u63d0\u4f9b\u65b9<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u5411\u91cf\u7ef4\u5ea6<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u6700\u5927\u8f93\u5165 token<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u4e2d\u6587\u6548\u679c<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u90e8\u7f72\u65b9\u5f0f<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u5907\u6ce8<\/th><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">text-embedding-3-small<\/td><td class=\"has-text-align-left\" data-align=\"left\">OpenAI<\/td><td class=\"has-text-align-left\" data-align=\"left\">1536<\/td><td class=\"has-text-align-left\" data-align=\"left\">8191<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4e2d\u7b49<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4ec5\u4e91\u7aef API<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u6027\u4ef7\u6bd4\u9ad8\uff0c\u9002\u5408\u82f1\u6587\u4e3a\u4e3b\u7684\u573a\u666f<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">text-embedding-3-large<\/td><td class=\"has-text-align-left\" data-align=\"left\">OpenAI<\/td><td class=\"has-text-align-left\" data-align=\"left\">3072<\/td><td class=\"has-text-align-left\" data-align=\"left\">8191<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4e2d\u7b49<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4ec5\u4e91\u7aef API<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u7ef4\u5ea6\u66f4\u9ad8\uff0c\u6548\u679c\u66f4\u597d\uff0c\u6210\u672c\u4e5f\u66f4\u9ad8<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">text-embedding-v3<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u963f\u91cc\u901a\u4e49<\/td><td class=\"has-text-align-left\" data-align=\"left\">1024\/768<\/td><td class=\"has-text-align-left\" data-align=\"left\">8192<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f18\u79c0<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4e91\u7aef API<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4e2d\u6587\u6548\u679c\u597d\uff0c\u652f\u6301\u591a\u79cd\u7ef4\u5ea6\u8f93\u51fa<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">BGE-large-zh<\/td><td class=\"has-text-align-left\" data-align=\"left\">BAAI\uff08\u667a\u6e90\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">1024<\/td><td class=\"has-text-align-left\" data-align=\"left\">512<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f18\u79c0<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u672c\u5730\u90e8\u7f72\/API<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5f00\u6e90\u6a21\u578b\uff0c\u4e2d\u6587\u6548\u679c\u7a81\u51fa<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">BGE-M3<\/td><td class=\"has-text-align-left\" data-align=\"left\">BAAI\uff08\u667a\u6e90\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">1024<\/td><td class=\"has-text-align-left\" data-align=\"left\">8192<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f18\u79c0<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u672c\u5730\u90e8\u7f72\/API<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u652f\u6301\u591a\u8bed\u8a00\u3001\u591a\u7c92\u5ea6\uff0c\u7efc\u5408\u80fd\u529b\u5f3a<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">Qwen3-Embedding-8B<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u963f\u91cc\u901a\u4e49<\/td><td class=\"has-text-align-left\" data-align=\"left\">4096<\/td><td class=\"has-text-align-left\" data-align=\"left\">32768<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f18\u79c0<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u672c\u5730\u90e8\u7f72\/API<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u6700\u65b0\u4e00\u4ee3\uff0c\u7ef4\u5ea6\u9ad8\uff0c\u4e0a\u4e0b\u6587\u7a97\u53e3\u5927<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">GTE-large-zh<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u963f\u91cc\u901a\u4e49<\/td><td class=\"has-text-align-left\" data-align=\"left\">1024<\/td><td class=\"has-text-align-left\" data-align=\"left\">8192<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4f18\u79c0<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u672c\u5730\u90e8\u7f72\/API<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u4e2d\u6587\u57fa\u51c6\u6d4b\u8bd5\u8868\u73b0\u597d<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<p>\u7ef4\u5ea6\u662f\u9009\u6a21\u578b\u65f6\u7ed1\u5b9a\u7684\u2014\u2014\u4f60\u9009\u4e86\u67d0\u4e2a\u6a21\u578b\uff0c\u7ef4\u5ea6\u5c31\u786e\u5b9a\u4e86\uff08\u90e8\u5206\u6a21\u578b\u652f\u6301\u591a\u79cd\u7ef4\u5ea6\u8f93\u51fa\uff0c\u4f46\u5927\u591a\u6570\u662f\u56fa\u5b9a\u7684\uff09\u3002<\/p>\n\n\n\n<p>\u6253\u4e2a\u6bd4\u65b9\uff1a\u7ef4\u5ea6\u5c31\u50cf\u63cf\u8ff0\u4e00\u4e2a\u4eba\u7528\u4e86\u591a\u5c11\u4e2a\u7279\u5f81\u3002\u7528 2 \u4e2a\u7279\u5f81\uff08\u8eab\u9ad8\u3001\u4f53\u91cd\uff09\u63cf\u8ff0\u4e00\u4e2a\u4eba\uff0c\u4fe1\u606f\u5f88\u6709\u9650\uff0c\u5f88\u591a\u4eba\u4f1a\u201c\u649e\u886b\u201d\u3002\u7528 100 \u4e2a\u7279\u5f81\uff08\u8eab\u9ad8\u3001\u4f53\u91cd\u3001\u80a4\u8272\u3001\u53d1\u578b\u3001\u53e3\u97f3\u3001\u8d70\u8def\u59ff\u52bf\u2026\u2026\uff09\u63cf\u8ff0\uff0c\u533a\u5206\u5ea6\u5c31\u9ad8\u591a\u4e86\u3002<strong>\u4f46\u7279\u5f81\u8d8a\u591a\uff0c\u8bb0\u5f55\u548c\u6bd4\u8f83\u7684\u6210\u672c\u4e5f\u8d8a\u9ad8\u3002<\/strong><\/p>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><tbody><tr><th class=\"has-text-align-left\" data-align=\"left\">\u7ef4\u5ea6\u8303\u56f4<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u9002\u7528\u573a\u666f<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u5b58\u50a8\u6210\u672c\uff08100 \u4e07\u6761\uff09<\/th><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">256~512<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u7b80\u5355\u573a\u666f\uff0c\u6587\u672c\u77ed\u3001\u7c7b\u76ee\u5c11<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u7ea6 1~2 GB<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">768~1024<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5927\u591a\u6570\u751f\u4ea7\u573a\u666f\u7684\u751c\u871c\u70b9<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u7ea6 3~4 GB<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">1536~4096<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5bf9\u7cbe\u5ea6\u8981\u6c42\u6781\u9ad8\u7684\u573a\u666f<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u7ea6 6~16 GB<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>\u5bf9\u4e8e\u5927\u591a\u6570\u4e2d\u6587 RAG \u9879\u76ee\uff0c768 \u5230 1024 \u7ef4\u662f\u4e00\u4e2a\u6bd4\u8f83\u7a33\u59a5\u7684\u9009\u62e9\u3002\u65e2\u80fd\u4fdd\u8bc1\u8db3\u591f\u7684\u8bed\u4e49\u533a\u5206\u5ea6\uff0c\u5b58\u50a8\u548c\u68c0\u7d22\u6210\u672c\u4e5f\u5728\u53ef\u63a7\u8303\u56f4\u5185\u3002\u9664\u975e\u4f60\u7684\u573a\u666f\u5bf9\u7cbe\u5ea6\u6709\u6781\u81f4\u8981\u6c42\uff08\u6bd4\u5982\u6cd5\u5f8b\u6761\u6587\u68c0\u7d22\u3001\u533b\u7597\u77e5\u8bc6\u5e93\u7b49\uff09\uff0c\u5426\u5219\u4e0d\u9700\u8981\u4e0a 3072 \u6216 4096 \u7ef4\u3002<\/p>\n<\/blockquote>\n\n\n\n<h2 class=\"wp-block-heading\">\u76f8\u4f3c\u5ea6\u8ba1\u7b97\uff1a\u600e\u4e48\u5224\u65ad\u4e24\u4e2a\u5411\u91cf\u201c\u50cf\u4e0d\u50cf\u201d<\/h2>\n\n\n\n<p>\u6587\u672c\u53d8\u6210\u5411\u91cf\u4e4b\u540e\uff0c\u4e0b\u4e00\u6b65\u5c31\u662f\u6bd4\u8f83\u4e24\u4e2a\u5411\u91cf\u4e4b\u95f4\u7684\u76f8\u4f3c\u7a0b\u5ea6\u3002\u7528\u6237\u8f93\u5165\u4e00\u4e2a query\uff0c\u7cfb\u7edf\u628a\u5b83\u8f6c\u6210\u5411\u91cf\uff0c\u7136\u540e\u548c\u77e5\u8bc6\u5e93\u91cc\u6240\u6709 chunk \u7684\u5411\u91cf\u9010\u4e00\u6bd4\u8f83\uff0c\u627e\u51fa\u6700\u76f8\u4f3c\u7684\u51e0\u4e2a\u2014\u2014\u8fd9\u5c31\u662f\u8bed\u4e49\u68c0\u7d22\u7684\u6838\u5fc3\u6d41\u7a0b\u3002<\/p>\n\n\n\n<p>\u90a3\u600e\u4e48\u6bd4\u8f83\u4e24\u4e2a\u5411\u91cf\u201c\u50cf\u4e0d\u50cf\u201d\uff1f\u8fd9\u5c31\u6d89\u53ca\u5230\u76f8\u4f3c\u5ea6\u8ba1\u7b97\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"1\">1. \u4f59\u5f26\u76f8\u4f3c\u5ea6\u2014\u2014\u6700\u5e38\u7528\u7684\u5ea6\u91cf\u65b9\u5f0f<\/h3>\n\n\n\n<p>\u4f59\u5f26\u76f8\u4f3c\u5ea6\uff08Cosine Similarity\uff09\u662f Embedding \u68c0\u7d22\u4e2d\u6700\u5e38\u7528\u7684\u76f8\u4f3c\u5ea6\u5ea6\u91cf\u65b9\u5f0f\u3002<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li> \u7b97\u4e24\u4e2a\u5411\u91cf\u7684\u70b9\u79ef\uff08\u5bf9\u5e94\u4f4d\u7f6e\u7684\u6570\u5b57\u76f8\u4e58\uff0c\u7136\u540e\u5168\u90e8\u52a0\u8d77\u6765\uff09<\/li>\n\n\n\n<li> \u7b97\u6bcf\u4e2a\u5411\u91cf\u7684\u6a21\uff08\u6bcf\u4e2a\u6570\u5b57\u7684\u5e73\u65b9\u52a0\u8d77\u6765\uff0c\u518d\u5f00\u6839\u53f7\uff09<\/li>\n\n\n\n<li> \u70b9\u79ef\u9664\u4ee5\u4e24\u4e2a\u6a21\u7684\u4e58\u79ef<\/li>\n<\/ol>\n\n\n\n<p>\u5728\u5b9e\u9645\u7684 RAG \u7cfb\u7edf\u4e2d\uff0c\u901a\u5e38\u4f1a\u8bbe\u4e00\u4e2a\u76f8\u4f3c\u5ea6\u9608\u503c\uff08threshold\uff09\uff0c\u53ea\u8fd4\u56de\u76f8\u4f3c\u5ea6\u9ad8\u4e8e\u9608\u503c\u7684\u7ed3\u679c\u3002<\/p>\n\n\n\n<p>\u5e38\u89c1\u7684\u505a\u6cd5\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u9608\u503c\u8bbe 0.7\uff1a\u6bd4\u8f83\u4e25\u683c\uff0c\u53ea\u8fd4\u56de\u9ad8\u5ea6\u76f8\u5173\u7684\u7ed3\u679c\uff0c\u51c6\u786e\u7387\u9ad8\u4f46\u53ef\u80fd\u6f0f\u6389\u4e00\u4e9b\u76f8\u5173\u5185\u5bb9<\/li>\n\n\n\n<li>\u9608\u503c\u8bbe 0.5\uff1a\u6bd4\u8f83\u5bbd\u677e\uff0c\u53ec\u56de\u7387\u9ad8\u4f46\u53ef\u80fd\u6df7\u5165\u4e00\u4e9b\u4e0d\u592a\u76f8\u5173\u7684\u5185\u5bb9<\/li>\n\n\n\n<li>\u4e0d\u8bbe\u9608\u503c\uff0c\u53ea\u53d6 Top-K\uff1a\u8fd4\u56de\u76f8\u4f3c\u5ea6\u6700\u9ad8\u7684 K \u4e2a\u7ed3\u679c\uff08\u6bd4\u5982 Top-5\uff09\uff0c\u4e0d\u7ba1\u5206\u6570\u591a\u5c11<\/li>\n<\/ul>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>\u5b9e\u9645\u9879\u76ee\u4e2d\uff0c\u5efa\u8bae\u5148\u7528 Top-K\uff08\u6bd4\u5982 K=5\uff09+ \u9608\u503c 0.6 \u7684\u7ec4\u5408\u7b56\u7565\uff1a\u5148\u53d6\u76f8\u4f3c\u5ea6\u6700\u9ad8\u7684 5 \u4e2a\uff0c\u518d\u8fc7\u6ee4\u6389\u4f4e\u4e8e 0.6 \u7684\u3002\u5177\u4f53\u9608\u503c\u9700\u8981\u6839\u636e\u4f60\u7684\u6570\u636e\u548c\u573a\u666f\u8c03\u8bd5\uff0c\u6ca1\u6709\u653e\u4e4b\u56db\u6d77\u800c\u7686\u51c6\u7684\u6570\u5b57\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>\u540c\u6837\u4e24\u6bb5\u6587\u672c\uff0c\u7528\u4e0d\u540c\u7684\u6a21\u578b\u7b97\u51fa\u6765\u7684\u76f8\u4f3c\u5ea6\u5206\u6570\u53ef\u80fd\u5dee\u5f88\u591a\u3002\u8fd9\u662f\u56e0\u4e3a\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4e0d\u540c\u6a21\u578b\u7684\u8bad\u7ec3\u6570\u636e\u4e0d\u540c\uff0c\u5bf9\u8bed\u4e49\u7684\u7406\u89e3\u65b9\u5f0f\u4e0d\u540c<\/li>\n\n\n\n<li>\u4e0d\u540c\u6a21\u578b\u7684\u5411\u91cf\u7ef4\u5ea6\u4e0d\u540c\uff0c\u8868\u8fbe\u80fd\u529b\u4e0d\u540c<\/li>\n\n\n\n<li>\u6709\u4e9b\u6a21\u578b\u8f93\u51fa\u7684\u5411\u91cf\u5df2\u7ecf\u505a\u4e86\u5f52\u4e00\u5316\uff08\u957f\u5ea6\u4e3a 1\uff09\uff0c\u6709\u4e9b\u6ca1\u6709<\/li>\n<\/ul>\n\n\n\n<p>\u6240\u4ee5\uff0c\u76f8\u4f3c\u5ea6\u9608\u503c\u4e0d\u80fd\u8de8\u6a21\u578b\u5957\u7528\u3002\u6362\u4e86\u6a21\u578b\u4e4b\u540e\uff0c\u9608\u503c\u8981\u91cd\u65b0\u8c03\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"610\" src=\"https:\/\/eve2333.top\/wp-content\/uploads\/2026\/03\/1774871236-image-1024x610.png\" alt=\"\" class=\"wp-image-1873\" srcset=\"https:\/\/eve2333.top\/wp-content\/uploads\/2026\/03\/1774871236-image-1024x610.png 1024w, https:\/\/eve2333.top\/wp-content\/uploads\/2026\/03\/1774871236-image-300x179.png 300w, https:\/\/eve2333.top\/wp-content\/uploads\/2026\/03\/1774871236-image-768x458.png 768w, https:\/\/eve2333.top\/wp-content\/uploads\/2026\/03\/1774871236-image-1536x915.png 1536w, https:\/\/eve2333.top\/wp-content\/uploads\/2026\/03\/1774871236-image-2048x1220.png 2048w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>demo \u91cc\u6211\u4eec\u4e00\u6b21\u6027\u628a 5 \u4e2a chunk \u4f20\u7ed9 API\uff0c\u5b9e\u9645\u9879\u76ee\u4e2d\u53ef\u80fd\u6709\u51e0\u4e07\u751a\u81f3\u51e0\u5341\u4e07\u4e2a chunk\u3002\u4e00\u6b21\u5168\u4f20\u8fc7\u53bb\u4e0d\u73b0\u5b9e\uff08API \u6709\u8bf7\u6c42\u5927\u5c0f\u9650\u5236\uff09\uff0c\u4e00\u4e2a\u4e00\u4e2a\u4f20\u53c8\u592a\u6162\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"2-1\">2.1 \u5206\u6279\u5904\u7406<\/h4>\n\n\n\n<p>\u6700\u57fa\u672c\u7684\u4f18\u5316\uff1a\u628a chunks \u5206\u6210\u56fa\u5b9a\u5927\u5c0f\u7684\u6279\u6b21\uff0c\u9010\u6279\u8c03\u7528 API\u3002<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>\/**\n * \u5206\u6279\u5411\u91cf\u5316\n *\n * @param texts     \u6240\u6709\u5f85\u5411\u91cf\u5316\u7684\u6587\u672c\n * @param batchSize \u6bcf\u6279\u7684\u5927\u5c0f\uff08\u5efa\u8bae 20~50\uff09\n * @return \u6240\u6709\u6587\u672c\u7684\u5411\u91cf\n *\/publicList&lt;double&#91;]&gt;embedInBatches(List&lt;String&gt; texts,int batchSize)throwsException{List&lt;double&#91;]&gt; allEmbeddings =newArrayList&lt;&gt;();for(int i =0; i &lt; texts.size(); i += batchSize){int end =Math.min(i + batchSize, texts.size());List&lt;String&gt; batch = texts.subList(i, end);System.out.printf(\"\u5411\u91cf\u5316\u8fdb\u5ea6\uff1a%d\/%d%n\", end, texts.size());List&lt;double&#91;]&gt; batchEmbeddings =embed(batch);\n        allEmbeddings.addAll(batchEmbeddings);\/\/ \u7b80\u5355\u7684\u9650\u6d41\uff1a\u6bcf\u6279\u4e4b\u95f4\u7b49\u4e00\u4e0b\uff0c\u907f\u514d\u89e6\u53d1 API \u7684\u901f\u7387\u9650\u5236if(end &lt; texts.size()){Thread.sleep(200);}}return allEmbeddings;}<\/code><\/pre>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"2-2\">2.2 \u5e76\u53d1\u63a7\u5236<\/h4>\n\n\n\n<p>\u5206\u6279\u5904\u7406\u662f\u4e32\u884c\u7684\uff0c\u5982\u679c API \u652f\u6301\u5e76\u53d1\uff0c\u53ef\u4ee5\u7528\u591a\u7ebf\u7a0b\u52a0\u901f\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>importjava.util.concurrent.*;\/**\n * \u5e76\u53d1\u6279\u91cf\u5411\u91cf\u5316\n *\/publicList&lt;double&#91;]&gt;embedConcurrently(List&lt;String&gt; texts,int batchSize,int maxConcurrency)throwsException{ExecutorService executor =Executors.newFixedThreadPool(maxConcurrency);List&lt;Future&lt;List&lt;double&#91;]&gt;&gt;&gt; futures =newArrayList&lt;&gt;();for(int i =0; i &lt; texts.size(); i += batchSize){int start = i;int end =Math.min(i + batchSize, texts.size());List&lt;String&gt; batch = texts.subList(start, end);\n\n        futures.add(executor.submit(()-&gt;embed(batch)));}List&lt;double&#91;]&gt; allEmbeddings =newArrayList&lt;&gt;();for(Future&lt;List&lt;double&#91;]&gt;&gt; future : futures){\n        allEmbeddings.addAll(future.get());}\n\n    executor.shutdown();return allEmbeddings;}<\/code><\/pre>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>\u5e76\u53d1\u6570\u4e0d\u8981\u8bbe\u592a\u9ad8\uff0c\u4e00\u822c 3~5 \u5c31\u591f\u4e86\u3002\u8bbe\u592a\u9ad8\u5bb9\u6613\u89e6\u53d1 API \u7684\u901f\u7387\u9650\u5236\uff08Rate Limit\uff09\uff0c\u53cd\u800c\u66f4\u6162\u3002<\/p>\n<\/blockquote>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"2-3\">2.3 \u9519\u8bef\u91cd\u8bd5<\/h4>\n\n\n\n<p>\u7f51\u7edc\u8bf7\u6c42\u96be\u514d\u4f1a\u5931\u8d25\uff0c\u52a0\u4e00\u4e2a\u7b80\u5355\u7684\u91cd\u8bd5\u673a\u5236\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>\/**\n * \u5e26\u91cd\u8bd5\u7684 embed \u65b9\u6cd5\n *\/publicList&lt;double&#91;]&gt;embedWithRetry(List&lt;String&gt; texts,int maxRetries)throwsException{Exception lastException =null;for(int attempt =1; attempt &lt;= maxRetries; attempt++){try{returnembed(texts);}catch(Exception e){\n            lastException = e;System.err.printf(\"\u7b2c %d \u6b21\u8c03\u7528\u5931\u8d25\uff1a%s\uff0c%s%n\",\n                    attempt, e.getMessage(),\n                    attempt &lt; maxRetries ?\"\u51c6\u5907\u91cd\u8bd5...\":\"\u5df2\u8fbe\u6700\u5927\u91cd\u8bd5\u6b21\u6570\");if(attempt &lt; maxRetries){\/\/ \u6307\u6570\u9000\u907f\uff1a\u7b2c 1 \u6b21\u7b49 1 \u79d2\uff0c\u7b2c 2 \u6b21\u7b49 2 \u79d2\uff0c\u7b2c 3 \u6b21\u7b49 4 \u79d2Thread.sleep(1000L*(1&lt;&lt;(attempt -1)));}}}thrownewRuntimeException(\"\u5411\u91cf\u5316\u5931\u8d25\uff0c\u5df2\u91cd\u8bd5 \"+ maxRetries +\" \u6b21\", lastException);}<\/code><\/pre>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"3\">\u5411\u91cf\u5316\u548c\u5143\u6570\u636e\u7684\u5173\u7cfb<\/h3>\n\n\n\n<p>\u4e00\u53e5\u8bdd\u6982\u62ec\uff1a<strong>\u5143\u6570\u636e\u4e0d\u53c2\u4e0e\u5411\u91cf\u5316\uff0c\u4f46\u548c\u5411\u91cf\u4e00\u8d77\u5b58\u5165\u5411\u91cf\u6570\u636e\u5e93<\/strong>\u3002<\/p>\n\n\n\n<p>\u6253\u4e2a\u6bd4\u65b9\uff1a\u5411\u91cf\u5316\u5c31\u50cf\u7ed9\u4e00\u672c\u4e66\u62cd\u4e86\u4e00\u5f20\u201c\u8bed\u4e49\u7167\u7247\u201d\uff0c\u5143\u6570\u636e\u5c31\u662f\u8d34\u5728\u7167\u7247\u80cc\u9762\u7684\u6807\u7b7e\uff08\u4f5c\u8005\u3001\u51fa\u7248\u65e5\u671f\u3001\u5206\u7c7b\u7b49\uff09\u3002\u62cd\u7167\u7684\u65f6\u5019\u4e0d\u9700\u8981\u770b\u6807\u7b7e\uff0c\u4f46\u5b58\u6863\u7684\u65f6\u5019\u6807\u7b7e\u8981\u8ddf\u7167\u7247\u653e\u5728\u4e00\u8d77\u3002<\/p>\n\n\n\n<p>\u68c0\u7d22\u7684\u65f6\u5019\uff0c\u6d41\u7a0b\u662f\u8fd9\u6837\u7684\uff1a<\/p>\n\n\n\n<p>1.\u5411\u91cf\u7f16\u7801\u6587\u672c\u548c \u5411\u91cf<br>2.ANN \u68c0\u7d22Query\u5411\u91cf<br>\u76f8\u4f3c\u5ea6\u8ba1\u7b97 \uff0c\u8fd4\u56deTop-K\uff0c\u6309\u76f8\u4f3c\u5ea6\u964d\u5e8f\uff0c\u5019\u9009 chunks<br>3.\u5143\u6570\u636e\u8fc7\u6ee4\u6839\u636e\u6761\u4ef6\u5f97\u5230\u6700\u7ec8\u7ed3\u679c<\/p>\n\n\n\n<p>\u6240\u4ee5\u5728\u5b58\u50a8\u9636\u6bb5\uff0c\u6bcf\u4e2a chunk \u5728\u5411\u91cf\u6570\u636e\u5e93\u91cc\u7684\u8bb0\u5f55\u901a\u5e38\u5305\u542b\u4e09\u90e8\u5206\uff1a<\/p>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><tbody><tr><th class=\"has-text-align-left\" data-align=\"left\">\u5b57\u6bb5<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u5185\u5bb9<\/th><th class=\"has-text-align-left\" data-align=\"left\">\u8bf4\u660e<\/th><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">id<\/td><td class=\"has-text-align-left\" data-align=\"left\">chunk \u7684\u552f\u4e00\u6807\u8bc6<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u7528\u4e8e\u66f4\u65b0\u548c\u5220\u9664<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">vector<\/td><td class=\"has-text-align-left\" data-align=\"left\">Embedding \u6a21\u578b\u8f93\u51fa\u7684\u5411\u91cf<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u7528\u4e8e\u76f8\u4f3c\u5ea6\u68c0\u7d22<\/td><\/tr><tr><td class=\"has-text-align-left\" data-align=\"left\">metadata<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u5143\u6570\u636e\uff08JSON \u683c\u5f0f\uff09<\/td><td class=\"has-text-align-left\" data-align=\"left\">\u7528\u4e8e\u8fc7\u6ee4\u548c\u5c55\u793a<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<p>\u5927\u90e8\u5206\u5411\u91cf\u6570\u636e\u5e93\uff08\u5982 Milvus\uff09\u8fd8\u53ef\u4ee5\u628a\u539f\u59cb\u6587\u672c\u4e5f\u5b58\u8fdb\u53bb\uff0c\u8fd9\u6837\u68c0\u7d22\u7684\u65f6\u5019\u4e0d\u7528\u518d\u53bb\u522b\u7684\u5730\u65b9\u53d6\u6587\u672c\u5185\u5bb9\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">\u4ec0\u4e48\u65f6\u5019\u9700\u8981\u91cd\u65b0\u5411\u91cf\u5316<\/h3>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"4-1-embedding\">4.1 \u6362\u4e86 Embedding \u6a21\u578b<\/h4>\n\n\n\n<p>\u4e0d\u540c\u6a21\u578b\u751f\u6210\u7684\u5411\u91cf\u662f\u4e0d\u517c\u5bb9\u7684\u3002\u6a21\u578b A \u751f\u6210\u7684\u5411\u91cf\u548c\u6a21\u578b B \u751f\u6210\u7684\u5411\u91cf\uff0c\u7ef4\u5ea6\u53ef\u80fd\u4e0d\u540c\uff0c\u5373\u4f7f\u7ef4\u5ea6\u76f8\u540c\uff0c\u8bed\u4e49\u7a7a\u95f4\u4e5f\u4e0d\u4e00\u6837\uff0c\u4e0d\u80fd\u6df7\u5728\u4e00\u8d77\u505a\u76f8\u4f3c\u5ea6\u8ba1\u7b97\u3002<\/p>\n\n\n\n<p>\u6362\u6a21\u578b = \u6240\u6709 chunk \u91cd\u65b0\u5411\u91cf\u5316 + \u5411\u91cf\u6570\u636e\u5e93\u91cc\u7684\u65e7\u5411\u91cf\u5168\u90e8\u66ff\u6362\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"4-2\">4.2 \u6587\u6863\u5185\u5bb9\u66f4\u65b0<\/h4>\n\n\n\n<p>\u5982\u679c\u67d0\u4efd\u6587\u6863\u7684\u5185\u5bb9\u6539\u4e86\uff08\u6bd4\u5982\u9000\u8d27\u653f\u7b56\u4ece 7 \u5929\u6539\u6210\u4e86 15 \u5929\uff09\uff0c\u5bf9\u5e94\u7684 chunk \u6587\u672c\u53d8\u4e86\uff0c\u5411\u91cf\u81ea\u7136\u4e5f\u8981\u91cd\u65b0\u751f\u6210\u3002<\/p>\n\n\n\n<p>\u5b9e\u9645\u64cd\u4f5c\u4e2d\uff0c\u901a\u5e38\u7684\u505a\u6cd5\u662f\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>1.\u6839\u636e <code>doc_id<\/code> \u627e\u5230\u8fd9\u4efd\u6587\u6863\u7684\u6240\u6709\u65e7 chunk<\/li>\n\n\n\n<li>2.\u5220\u9664\u65e7 chunk \u7684\u5411\u91cf<\/li>\n\n\n\n<li>3.\u5bf9\u66f4\u65b0\u540e\u7684\u6587\u6863\u91cd\u65b0\u5206\u5757\u3001\u5411\u91cf\u5316\u3001\u5b58\u5165\u5411\u91cf\u6570\u636e\u5e93<\/li>\n<\/ol>\n\n\n\n<p>\u8fd9\u5c31\u662f\u4e3a\u4ec0\u4e48\u524d\u4e00\u7bc7\u5f3a\u8c03\u5143\u6570\u636e\u91cc\u8981\u6709 <code>doc_id<\/code>\u2014\u2014\u6ca1\u6709\u5b83\uff0c\u4f60\u5f88\u96be\u77e5\u9053\u54ea\u4e9b chunk \u5c5e\u4e8e\u540c\u4e00\u4efd\u6587\u6863\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"4-3\">4.3 \u5206\u5757\u7b56\u7565\u8c03\u6574<\/h4>\n\n\n\n<p>\u5982\u679c\u4f60\u8c03\u6574\u4e86\u5206\u5757\u7684\u5927\u5c0f\u6216\u91cd\u53e0\u7b56\u7565\uff08\u6bd4\u5982\u4ece 500 \u5b57\u4e00\u5757\u6539\u6210 300 \u5b57\u4e00\u5757\uff09\uff0cchunk \u7684\u5185\u5bb9\u53d8\u4e86\uff0c\u5411\u91cf\u4e5f\u8981\u91cd\u65b0\u751f\u6210\u3002<\/p>\n\n\n\n<p>\u8fd9\u79cd\u60c5\u51b5\u901a\u5e38\u610f\u5473\u7740\u5168\u91cf\u91cd\u65b0\u5411\u91cf\u5316\uff0c\u5de5\u4f5c\u91cf\u6bd4\u8f83\u5927\u3002\u6240\u4ee5\u5206\u5757\u7b56\u7565\u6700\u597d\u5728\u9879\u76ee\u521d\u671f\u5c31\u786e\u5b9a\u4e0b\u6765\uff0c\u522b\u9891\u7e41\u6539\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"4-4\">4.4 \u6a21\u578b\u5347\u7ea7<\/h4>\n\n\n\n<p>\u540c\u4e00\u4e2a\u6a21\u578b\u7684\u4e0d\u540c\u7248\u672c\uff08\u6bd4\u5982 <code>text-embedding-v2<\/code> \u5347\u7ea7\u5230 <code>text-embedding-v3<\/code>\uff09\uff0c\u751f\u6210\u7684\u5411\u91cf\u4e5f\u53ef\u80fd\u4e0d\u517c\u5bb9\u3002\u5347\u7ea7\u524d\u8981\u770b\u6a21\u578b\u63d0\u4f9b\u65b9\u7684\u8bf4\u660e\uff0c\u786e\u8ba4\u65b0\u65e7\u7248\u672c\u7684\u5411\u91cf\u662f\u5426\u517c\u5bb9\u3002<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>\u4e00\u4e2a\u5b9e\u7528\u7684\u5efa\u8bae\uff1a\u5728\u5411\u91cf\u6570\u636e\u5e93\u7684\u5143\u6570\u636e\u91cc\u8bb0\u5f55 <code>embedding_model<\/code> \u548c <code>embedding_model_version<\/code>\uff0c\u8fd9\u6837\u4f60\u968f\u65f6\u77e5\u9053\u6bcf\u4e2a\u5411\u91cf\u662f\u7528\u54ea\u4e2a\u6a21\u578b\u751f\u6210\u7684\uff0c\u5347\u7ea7\u65f6\u4e5f\u65b9\u4fbf\u505a\u7070\u5ea6\u5207\u6362\u3002<\/p>\n<\/blockquote>\n","protected":false},"excerpt":{"rendered":"<p>\u5143\u6570\u636e\u7684\u4f5c\u7528\u4e0e\u7ba1\u7406 \u5143\u6570\u636e\u662f\u5728\u5206\u5757\u4e4b\u540e\u3001\u5411\u91cf\u5316\u4e4b\u524d\u52a0\u5165\u7684\u3002\u5206\u5757\u5b8c\u6210\u540e\uff0c\u4f60\u5f97\u5230\u7684\u662f\u4e00\u4e2a\u4e2a\u7eaf\u6587\u672c\u5757\uff0c\u8fd9\u65f6\u5019\u7ed9\u6bcf\u4e2a\u5757\u6253\u4e0a\u6807\u7b7e\uff0c\u8bb0\u5f55\u5b83\u7684\u6765 &#8230;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"emotion":"","emotion_color":"","title_style":"","license":"","footnotes":""},"categories":[1],"tags":[],"class_list":["post-1869","post","type-post","status-publish","format-standard","hentry","category-uncategorized"],"_links":{"self":[{"href":"https:\/\/eve2333.top\/index.php?rest_route=\/wp\/v2\/posts\/1869","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/eve2333.top\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/eve2333.top\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/eve2333.top\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/eve2333.top\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=1869"}],"version-history":[{"count":3,"href":"https:\/\/eve2333.top\/index.php?rest_route=\/wp\/v2\/posts\/1869\/revisions"}],"predecessor-version":[{"id":1874,"href":"https:\/\/eve2333.top\/index.php?rest_route=\/wp\/v2\/posts\/1869\/revisions\/1874"}],"wp:attachment":[{"href":"https:\/\/eve2333.top\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1869"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/eve2333.top\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=1869"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/eve2333.top\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=1869"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}